We did indeed examine these issues in comparing the patient samples randomised to treatment and control conditions at study entry. They did not differ in life events on the Holmes-Rahe measure in maladaptive coping responses, 3 in the quality of social support as measured by the family environment sealer or in affect, including measures of anxiety, depression, and anger. 6 They differed slightly in socioeconomic status (SES) on the basis of husband’s occupation, 3, 7 but both groups had modal scores placing them in the same social class (II) on the Hollingshead-Redlich scale. 8 The between-group differences in survival time remained highly significant even when SES was entered first as a control variable in the proportional hazards model9 described in our study. This was true both for time from randomisation to death (n= 86, p< 0’0001) and time from first metastasis to death (n= 86, p< 0-009).

Chaturvedi then inquires whether there were subsequent differences in social or marital support. We have no evidence that this was so. Scores on the family environment scale did not differ between intervention and control groups at the end of the initial year of the study. However, even if there were other differences, they would simply suggest a plausible mechanism by which the group treatment may have asserted its influence. Our groups did indeed